The discovery of new molecules has traditionally focused in two broad areas, biologically active molecules, which are used as drugs for the treatment of life-threatening diseases, and new materials, which are used in commercial, and especially, in high technological applications. In both areas, the strategy used to discover new molecules has involved two basic operations: (i) a more or less random choice of a molecular candidate, prepared either via chemical synthesis or isolated from natural sources, and (ii) the testing of the molecular candidate for the property or properties of interest. This discovery cycle is repeated indefinitely until a molecule possessing the desirable property “lead molecule”, is located. This “lead molecule” discovery process has been inherently ad-hoc in nature and is time-consuming, laborious, unpredictable and costly.
Once a candidate “lead” molecule has been located, the synthetic chemist must subsequently find ways to synthesize structural variants of this lead molecule to optimize its properties in the desired application. In the case where the “lead” molecule is a synthetized organic species or a natural product, the chemist is usually limited to certain structural themes and synthetic reaction schemes. These are dictated largely by the structural composition of the “lead” molecule and by the requirements of the specific application. For example, in cases where the “lead” posesses a functionally important aromatic ring, various electrophillic and nucleophillic substitutions are typically be carried out on the ring to produce variants. Each such case must be approached as a specific independent design and synthesis problem, starting each time from the beginning, because of the lack of availability of an appropriate chemistry to simply alter the structure of the lead compound to produce the variant.
Recently, some attempts have been made to modularize certain synthetic organic reaction schemes to facillitate modification and transformation of a base compound (see, for example, Proc. Natl. Acad. Sci. USA, 90, 6909, 1933). However, the molecules which can be produced by such attempts are extremely limited in their achievable diversity and are still bounded by factors dictated by the choice of specific structural themes. In the case where the “lead molecule” is a naturally occuring biological molecule, such as a peptide, a protein, an oligonucleotide or a carbohydrate, simple synthetic point-modifications to the lead molecule to produce variants are quite difficult to achieve.
A brief account of the strategies and tactics used in the discovery of new molecules is described below. The emphasis is on biologically interesting molecules; however, the technical problems encountered in the discovery of biologically active molecules as outlined here are also illustrative of the problems encountered in the discovery of molecules which can serve as building blocks for the development of new tools and materials for a variety of high technological applications. Furthermore, as discussed below, these problems are also illustrative of the problems encountered in the development of fabricated structures and materials for high technological applications.
Drug Design
Modern theories of biological activity state that biological activities, and therefore physiological states, are the result of molecular recognition events. For example, nucleotides can form complementary base pairs so that complementary single-stranded molecules hybridize resulting in double- or triple-helical structures that appear to be involved in regulation of gene expression. In another example, a biologically active molecule, referred to as a ligand, binds with another molecule, usually a macromolecule referred to as ligand-acceptor (e.g., a receptor or an enzyme), and this binding elicits a chain of molecular events which ultimately gives rise to a physiological state, e.g., normal cell growth and differentiation, abnormal cell growth leading to carcinogenesis, blood-pressure regulation, nerve-impulse-generation and -propagation, etc. The binding between ligand and ligand-acceptor is geometrically characteristic and extraordinarily specific, involving appropriate three-dimensional structural arrangements and chemical interactions.
A currently favored strategy for the development of agents which can be used to treat diseases involves the discovery of forms of ligands of biological receptors, enzymes, or related macromolecules, which mimic such ligands and either boost, i.e., agonize, or suppress, i.e., antagonize, the activity of the ligand. The discovery of such desirable ligand forms has traditionally been carried out either by random screening of molecules (produced through chemical synthesis or isolated from natural sources), or by using a so-called “rational” approach involving identification of a lead-structure, usually the structure of the native ligand, and optimization of its properties through numerous cycles of structural redesign and biological testing. Since most useful drugs have been discovered not through the “rational” approach but through the screening of randomly chosen compounds, a hybrid approach to drug discovery has recently emerged which is based on the use of combinatorial chemistry to construct huge libraries of randomly-built chemical structures which are screened for specific biological activities. (S. Brenner and R. A. Lerner, 1992, Proc. Natl. Acad. Sci. USA 89:53, 81)
Most lead-structures which have been used in the “rational” drug design approach are native polypeptide ligands of receptors or enzymes. The majority of polypeptide ligands, especially the small ones, are relatively unstable in physiological fluids, due to the tendency of the peptide bond to undergo facile hydrolysis in acidic media or in the presence of peptidases. Thus, such ligands are decisively inferior in a pharmacokinetic sense to nonpeptidic compounds, and are not favored as drugs. An additional limitation of small peptides as drugs is their low affinity for ligand acceptors. This phenomenon is in sharp contrast to the affinity demonstrated by large, folded polypeptides, e.g., proteins, for specific acceptors, e.g., receptors or enzymes, which can be in the subnanomolar range. For peptides to become effective drugs, they must be transformed into nonpeptidic organic structures, i.e., peptide mimetics, which bind tightly, preferably in the nanomolar range, and can withstand the chemical and biochemical rigors of coexistence with biological tissues and fluids.
Despite numerous incremental advances in the art of peptidomimetic design, no general solution to the problem of converting a polypeptide-ligand structure to a peptidomimetic has been defined. At present, “rational” peptidomimetic design is done on an ad hoc basis. Using numerous redesign-synthesis-screening cycles, peptidic ligands belonging to a certain biochemical class have been converted by groups of organic chemists and pharmacologists to specific peptidomimetics; however, in the majority of cases results in one biochemical area, e.g., peptidase inhibitor design using the enzyme substrate as a lead, cannot be transferred for use in another area, e.g., tyrosine-kinase inhibitor design using the kinase substrate as a lead.
In many cases, the peptidomimetics that result from a peptide structural lead using the “rational” approach comprise unnatural alpha-amino acids. Many of these mimetics exhibit several of the troublesome features of native peptides (which also comprise alpha-amino acids) and are, thus, not favored for use as drugs. Recently, fundamental research on the use of nonpeptidic scaffolds, such as steroidal or sugar structures, to anchor specific receptor-binding groups in fixed geometric relationships have been described (see for example Hirschmann, R. et al., 1992 J. Am. Chem. Soc., 114:9699–9701; Hirschmann, R. et al., 1992 J. Am. Chem. Soc., 114:9217–9218); however, the success of this approach remains to be seen.
In an attempt to accelerate the identification of lead-structures, and also the identification of useful drug candidates through screening of randomly chosen compounds, researchers have developed automated methods for the generation of large combinatorial libraries of peptides and certain types of peptide mimetics, called “peptoids”, which are screened for a desirable biological activity. For example, the method of H. M. Geysen, (1984 Proc. Natl. Acad. Sci. USA 81:3998) employs a modification of Merrifield peptide synthesis, wherein the C-terminal amino acid residues of the peptides to be synthesized are linked to solid-support particles shaped as polyethylene pins; these pins are treated individually or collectively in sequence to introduce additional amino-acid residues forming the desired peptides. The peptides are then screened for activity without removing them from the pins. Houghton, (1985, Proc. Natl. Acad. Sci. USA 82:5131; and U.S. Pat. No. 4,631,211) utilizes individual polyethylene bags (“tea bags”) containing C-terminal amino acids bound to a solid support. These are mixed and coupled with the requisite amino acids using solid phase synthesis techniques. The peptides produced are then recovered and tested individually. Fodor et al., (1991, Science 251:767) described light-directed, spatially addressable parallel-peptide synthesis on a silicon wafer to generate large arrays of addressable peptides that can be directly tested for binding to biological targets. These workers have also developed recombinant DNA/genetic engineering methods for expressing huge peptide libraries on the surface of phages (Cwirla et al., 1990, Proc. Natl. Acad. Sci. USA 87:6378).
In another combinatorial approach, V. D. Huebner and D. V. Santi (U.S. Pat. No. 5,182,366) utilized functionalized polystyrene beads divided into portions each of which was acylated with a desired amino acid; the bead portions were mixed together, then divided into portions each of which was re-subjected to acylation with a second desirable amino acid producing dipeptides, using the techniques of solid phase peptide synthesis. By using this synthetic scheme, exponentially increasing numbers of peptides were produced in uniform which were then separately screened for a biological activity of interest.
Zuckerman et al., (1992, Int. J. Peptide Protein Res. 91:1) also have developed similar methods for the synthesis of peptide libraries and applied these methods to the automation of a modular synthetic chemistry for the production of libraries of N-alkyl glycine peptide derivatives, called “peptoids”, which are screened for activity against a variety of biochemical targets. (See also, Symon et al., 1992, Proc. Natl. Acad. Sci. USA 89:9367). Encoded combinatorial chemical syntheses have been described recently (S. Brenner and R. A. Lerner, 1992, Proc. Natl. Acad. Sci. USA 89:5381).
The focus of these structural diversity activities on peptide synthesis chemistry is a direct result of the fact that the ability to generate structural diversity requires, as its starting point, the access to practical stepwise sequential synthesis chemistries which allow the incorporation of varied structural elements with orthagonal reactivities. To-date, these have only been worked out for the Merrifield synthesis of peptides and the Carruthers synthesis of oligonucleotides. Thus, there remains a need for an improved method for the structure-directed generation and screening of organic compounds to determine which may be suitable in a particular application.